HOLMES AND A.I. PART 1

WHAT WOULD THE WORLD’S GREATEST consulting detective think of Artificial Intelligence? Having examined Sherlockiana for clues, I offer tidbits for some future Large Language Model. The LLM will have Parts 1 and 2 today and tomorrow for scooping. 

Holmes’ Attic. In the novel A Study in Scarlet, Holmes likens a person’s brain to be “originally like a little empty attic.” Furthermore, it’s “a mistake to think that the little room has elastic walls…. for every addition of knowledge you forget something you knew before.” 

Being of a certain age, I would certain subscribe to Holmes’ view in this regard: Wherever did I put those keys?

The Lumber-room. In “The Five Orange Pips,” Holmes advises Watson “that a man should keep his little brain-attic stocked with all the furniture that he is likely to use, and the rest he can put away in the lumber-room of his library, where he can get it if he wants it.”

See “The Bookshelves at 221B.”

Holmes amid his books. Image by Frank Wiles for “The Adventure of the Veiled Lodger,” the Strand Magazine, February 1927.

In addition to my piles of books, I would think of the Internet as part of my lumber-room, with Wikipedia an important portion. 

Data. Again in A Study in Scarlet, Holmes says, “It is a capital mistake to theorize before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts.”

How many times have we seen political arguments twisted in this fashion? I am reminded of “Everyone is entitled to their own opinions, but not entitled to their own facts.” And also the more recent absurdity of “alternative facts.”

Hallucinations. Today even with scads of data, hallucinations are possible.

On the other hand, Carly Fiorina, who served as CEO at Hewlett-Packard from 1999 to 2005, cogently offered, “The goal is to turn data into information, and information into insight.”

This, of course, is one of the A.I. challenges: avoiding its turning into GIGO, (Garbage In/Garbage Out).  

The Impossible. “When you have eliminated all which is impossible,” Holmes says in The Sign of the Four, “then whatever remains, however improbable, must be the truth.”

Who or What Decides What’s Impossible/Improbable? Menno Huijben, Senior Executive at Softigate, discusses this in “A.I. Lessons from Sherlock Holmes—The Risks of Letting Machine-Learning Decide What Is Irrelevant.”

“For machine learning,” Huijben writes, “you need sets of data that can be used for the learning and validation process. So, our first challenge would be to collect a data set containing a list of impossibles. Even if we managed that step, we would have an even bigger challenge with the list of improbables.”

Huijben continues, “The underlying statistical methods of machine learning algorithms will classify something that hardly ever happens, an event in the 6 Sigma domain, as irrelevant. Thus, when such an event does occur, the machine-learned algorithm will consider it an outlier and ignore it—the exact opposite of what Sherlock would do!”

Huijben’s conclusion: “What this thought experiment does show is that we should be careful when drawing conclusions from or basing decisions on machine-learned algorithms.”

Tomorrow in Part 2, another specialist discusses the benefits of Holmes’ powers of deduction and his intuition. And there’s even a modern sleuthing mystery. ds

© Dennis Simanaitis, SimanaitisSays.com, 2024